lmsys/vicuna-13b-v1.3

TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Jun 18, 2023Architecture:Transformer0.2K Cold

The lmsys/vicuna-13b-v1.3 is a 13 billion parameter chat assistant developed by LMSYS, fine-tuned from the LLaMA architecture with a 4096-token context length. It specializes in conversational AI, trained on user-shared conversations from ShareGPT. This model is primarily intended for research and hobbyist use in natural language processing and chatbot development.

Loading preview...

Overview

Vicuna v1.3 is a 13 billion parameter chat assistant developed by LMSYS, fine-tuned from the LLaMA model. It is designed for conversational AI tasks, leveraging a training dataset of approximately 125,000 user-shared conversations collected from ShareGPT.com. The model operates under a non-commercial license, making it suitable for research and hobbyist applications.

Key Capabilities

  • Conversational AI: Excels at generating human-like responses in chat-based interactions.
  • Instruction Following: Fine-tuned using supervised instruction fine-tuning, enhancing its ability to follow user prompts.
  • Research Tool: Primarily intended for researchers and hobbyists exploring large language models and chatbot development.

Evaluation and Performance

Vicuna's performance has been assessed using standard benchmarks, human preference studies, and an "LLM-as-a-judge" methodology. Further details on its evaluation and comparative performance can be found in its research paper and the LMSYS Chatbot Arena Leaderboard.

Getting Started

Users can interact with Vicuna via a command-line interface or through APIs compatible with OpenAI and Hugging Face. Detailed instructions are available in the FastChat repository.